Re: draft minutes from the sming interim:

[Andy Bierman <abierman@cisco.com>]

At 12:04 PM 6/28/2002, Frank Strauss wrote:
>Hi!
>
>David> [...] a well defined and standard annotation mechanism as
>David> part of the language is really very useful. There are several
>David> examples of the annotation that exist in commercial products
>David> such as the MIB compiler from Epilogue (WindRiver) to assist
>David> in generation of code for SNMP agents, and from Novell and HP
>David> to provide additional information about notifications.
>
>I agree.

I'm not disagreeing that MIB compiler writers may want to add
directives to MIB modules to assist in the generation of
output documents.  I'm concerned that such an effort is
inappropriate wrt/ the charter of the SMIng WG.  I'm concerned
that it will require a lengthy effort, interesting to a small number
of people, to agree on the syntax and semantics of these directives.
I'm concerned that these directives are not universally useful, but
rather specific to particular tools and vendors.  MIB readers
want to view MIBs in a consistent format.  The WG already
decided not to pursue compiler directives (in SLC).



>David> The three problems with ASN.1 comments are 
>David> 1) use of "--" to delimit comments
>David> 2) using the same chars to both start and terminate comments
>David> 3) comments cannot be continued over more than one line
>
>I agree just that (1) and (2) is a problem.

Juergen's proposal fixes 1 and 2.  It is not a burden on the
MIB reader or writer if the start-of-comment token is repeated
on each line. Also, this practice is more 'grep-friendly'.


>David> So, having both C-style comments of:
>David>    1) "/*" <comment_text> "*/" and
>David>    2) "//" <comment_text> <EOL>
>David> would be a welcome change.
>
>I don't think multi-line comments started by a single comment prefix
>on the first line, is not required. E.g., writing shell scripts is no
>problem for a couple of decades. However, I have no strong opinion
>against such multi-line comments.
>
>
>David> With SMICng there are many compiler directives to control
>David> checking.  Because there is no standard way to specify
>David> directives, and there are no standard directives, users of
>David> SMICng must specify the checking directives on the command line
>David> (or environment variable) to SMICng, or create wrapper files
>David> containing directives which then include the MIB file. I don't
>David> want to modify a MIB file so that it cannot be used by another
>David> compiler. This means that the directives are at the module
>David> level. It would be much better to specify the directive around
>David> the particular construct in a MIB module. This helps the reader
>David> of a MIB module by "pointing out" usage that is questionable
>David> (maybe for backwards compatibility), or providing a hint for
>David> unusual usage.
>
>If you would like to have a parser supporting directives within MIB
>files, this can be achieved within comments, without breaking SMI
>compatibility. (I'm not saying this is an elegant approach. But IMHO
>blowing the language with preprocessor stuff isn't any better.)

I agree with Frank. Putting directives outside the standard (in comments)
also prevents clever MIB writers from playing tricks with #if and #defines.



> -frank

Andy